Adaptive Neural Learning Finite-Time Control for Uncertain Teleoperation System with Output Constraints

Restricted by operation time and workspace, the end effector of robot needs to complete teleoperation tasks within the shortest possible time while satisfying physical constraints. To address the above issues, a novel adaptive neural learning finite-time control scheme with a modified funnel variable is developed. The developed approach considers the comprehensive characteristics of finite-time convergence and asymptotic convergence. Compared with the existing approach, prescribed performance and higher convergence rate can be guaranteed. In this study, neural networks are utilized to approximate various uncertainties, and a robust term is further employed to eliminate unknown external disturbances and estimation biases of neural networks. The previous passivity issue is avoided by replacing the signals transmitted on the cyber channel with virtual environmental parameters instead of high-frequency force signals. The transparency and task performance of the developed approach have been improved to a certain extent. Numerical simulations and experiments are conducted on a teleoperation platform consisting of a pair of Phantom Omni 3D Touch robots to validate the feasibility and availability.